This paper describes a study that used a computational tool called CMfinder to search for structured non-coding RNAs (ncRNAs) in vertebrate genomes within ENCODE regions. CMfinder identified over 6,500 candidate ncRNA regions with an estimated false positive rate of 50%. A subset of 11 candidates were experimentally verified, with 10 shown to be present as RNA transcripts in specific tissues. The study found that considering RNA secondary structure, as CMfinder does, can help identify ncRNAs that may be missed by sequence-based alignment methods alone. The study also discussed some of the challenges in accurately identifying ncRNAs computationally.